This video provides an engaging exploration into Seattle's booming real estate market and the transformative impact of technology in the industry. It opens with a humorous take on Amazon's architectural ventures in the city and swiftly delves into the high-stakes housing market, characterized by limited inventory and sky-high prices. A visit to Glen Kelman, CEO of Redfin, reveals insights into how the company aims to modernize the industry through tech, focusing on efficient, customer-friendly practices and addressing socio-economic disparities.

The video also highlights Redfin's innovative use of technology to navigate the volatile real estate market. By harnessing machine learning algorithms and data analysis, Redfin can identify trends and make informed decisions on property investments and renovations. This technology-driven approach enables the company to improve the buying and selling process, ensuring better returns and a fairer real estate landscape.

Main takeaways from the video:

Please remember to turn on the CC button to view the subtitles.

Key Vocabularies and Common Phrases:

1. sycophants [ˈsɪkəˌfænts] - (n.) - People who act obsequiously toward someone important in order to gain advantage. - Synonyms: (flatterers, yes-men, bootlickers)

Clearly no one could tell George Lucas what a racist disaster Jar Jar was because he was surrounded by 100, like, 18 layers of sycophants.

2. foil [fɔɪl] - (n.) - A person or thing that contrasts with and so emphasizes and enhances the qualities of another. - Synonyms: (contrast, complement, counter)

With real estate, we just found the perfect foil, where you have an agent who isn't always acting in the customer's best interest, but also charging a high price.

3. dubious [ˈduːbiəs] - (adj.) - Hesitating or doubting, not to be relied upon, suspect. - Synonyms: (doubtful, uncertain, questionable)

The promise of Redfin has long centered on lowering commission rates and trying to modernize real estate and some of its dubious practices.

4. affluent [ˈæfluənt] - (adj.) - Having a great deal of money; wealthy. - Synonyms: (wealthy, prosperous, rich)

Redfin, however, has faced recent criticism and even been sued for offering better services to affluent customers.

5. zoning [ˈzoʊnɪŋ] - (n.) - A system of land use regulation for public and private development. - Synonyms: (land division, land use planning, urban planning)

Vancouver just has different zoning than Seattle.

6. synthetic [sɪnˈθɛtɪk] - (adj.) - Made by chemical synthesis, especially to imitate a natural product. - Synonyms: (artificial, man-made, manufactured)

It makes something called synthetic data

7. procedural [prəˈsiːdʒərəl] - (adj.) - Relating to an established or official way of doing something. - Synonyms: (methodical, systematic, routine)

About not just training the AI, but training the tools that allow us to do this world building and procedural generation.

8. orbitrz ['ɔrbɪrts] - (n.) - The path of a celestial body or an artificial satellite as it revolves around another body. - Synonyms: (trajectory, flight path, course)

For example, startups are rushing to send satellites into orbitz that photograph the earth in different ways.

9. gaudy [ˈɡɔːdi] - (adj.) - Extravagantly bright or showy, typically so as to be tasteless. - Synonyms: (garish, flashy, ostentatious)

It's technically an Amazon office, but really, it's a gaudy showpiece.

10. revamp [riˈvæmp] - (v.) - Give new and improved form, structure, or appearance to. - Synonyms: (renovate, overhaul, refurbish)

Redfin has revamped the place in the hopes of kicking its price higher.

AI or Bust in Seattle's Real Estate Market

Walk downtown in Seattle, and you're likely to come across this. Oh, sorry. I mean, this. It's technically an Amazon office, but really, it's a gaudy showpiece. Look at us. We're Amazon. We built a rainforest smack dab in the middle of Seattle and encased it in silver balls because we can. It's the kind of thing that seems like a good idea right after you've drained your blood, boy, but that you later regret. Next thing you know, you have to build a stand right by your giant balls and give away free fruit all day as an apology. We're sorry for what we've done. Have as many bananas as you'd like.

Seattle's obscene wealth has turned the real estate market here into a brutal free for all. Just ask these people. The sprint to find permanent housing here. Whether you're buying or renting, is insane. And to actually buy a place feels out of the realm of possibility. Unless you're a millionaire. It just seems like it keeps getting worse and worse. The inventory is pretty tight. I didn't. It's hard to find really good stuff. I think renting was easier. I was kind of shocked at how expensive everything is and how competitive the market is. The Seattle housing scene, in one word, crazy. I'd say it's a gauntlet. Cutthroat. Trouble to learn about the latest in Seattle real estate and the art of selling a home.

Here, I went straight to the top. Let's roll. I paid a visit to my old friend Glen Kelman, the CEO of Redfin. I should warn you that almost no one is willing to get in the car with me while I'm driving. And if at any point, you need to grab the wheel to save us from death, I won't resist. About the yoke. It's terrible. Yeah, it's Elon musk. Just being opinionated. To be opinionated without actually being better. I have this theory, by the way. I think about when people can stand up to CEO's all the time. Cause I worry that people can't stand up to me. And for me, the signal version of that is Jar Jar Binks, where clearly no one could tell George Lucas what a racist disaster Jar Jar was because he was surrounded by 100, like, 18 layers of sycophants. I think no one could tell Elon, this is an accident waiting to happen, this steering wheel. And I can't figure out what mistakes Redfin is making that no one has the courage to tell me about.

I used Redfin back in. Hey, I didn't know that really? Yeah. I told you. No, I store that shit. I shouldn't have said the s word, but, no, you can't. No, you did, because I had gotten to know you, and I was like, oh, fine. I'll give this red fin thing a try then. I feel like you dated me when I was a teenager and had braces. It's so much better now. So technology is supposed to make industries more efficient. With real estate, we just found the perfect foil, where you have an agent who isn't always acting in the customer's best interest, but also charging a high price. And so we thought that by using technology, we could make the agent more efficient.

We're on our way to see a home that Redfin bought, refurbished, and then sold for a profit. Buying and flipping homes like this is a huge risk for the company. Just ask Zillow. If housing prices go down, Redfin might get stuck with money losing inventory. Glenn's belief is that Redfin can use its technology and experience to lower the risks. It analyzes thousands of data points to monitor quick changing trends in housing markets across the country. A mix of machine learning algorithms, and computer vision tell Redfin where to buy homes and what quick fixes will result in the best returns.

First of all, there's just the fundamental problem of trying to figure out what a home is worth. And the reason that you need machine learning algorithms rather than standard computer science is because a basement is very valuable in Seattle, whereas a pool is valuable in Atlanta. It is constantly re evaluating what factors matter the most. And so one of them is a pricing algorithm, and then another one is about renovations and computer vision and trying to assess the quality of the property based on photos. Oh, okay. The little house here used to look like this. Cozy, but not exactly Seattle's finest, unless you're into smoke tinged, yellowing walls. Redfin has revamped the place in the hopes of kicking its price higher. And Heather here is one of Redfin's chief revampers.

This isn't necessarily someone's individual personality, and, you know, we're just trying to position the house so people can see what the potential is. She's promised me some top tips, so we want to keep it kind of neutral. Are there any other, like, just tiny details that people wouldn't think of? Like, the kind of floss, I think, anything that you touch. Right. So, like, that's why we want to put in new plumbing fixtures, or we want to make sure, like, there's new knobs, because I think that people that makes it, you know, feel like, it's a good experience. Okay. Yeah. Now that makes sense. The simplest thing that has the highest ROI is just update the kitchen without replacing it. Okay. Paint is just transformative, so paint your whole house and keep it neutral. Makes sense. Make sure your flooring is consistent.

This might seem like normal renovation stuff, but it's all pulled from Redfin's algorithms. We're doing 5000 sales a year, and we're capturing data about what works and doesn't. And the simplest lesson is to not overdo it. When we have lost our shirt, it's almost always doing too much, not too little. How many times did you screw it up before you guys, you know, is this as if we stopped screwing it up?

The promise of Redfin has long centered on lowering commission rates and trying to modernize real estate and some of its dubious practices. Redfin, however, has faced recent criticism and even been sued for offering better services to affluent customers. Glenn says he takes these critiques to heart, and he promises that he's more than some rich guy capitalizing off a crazed real estate market.

People pay $60 to $80 billion in real estate commissions, and no one would ever pay that if you actually had to dig out your wallet and hand a real estate agent $30,000 or $40,000. So, technology is supposed to make industries more efficient. The second part of it is more about social justice, that still lenders who decide that someone who is latino or black is less likely to be able to afford a home, even if they got the same income. Like, why would your agents be less inclined to do those sorts of things than others? Historically, you can't control any human being in every way. But we can certainly say that when you list a home, you need to make sure that everybody can see that it's for sale instead of just posting it on Facebook, which has become especially popular in Silicon Valley, you give people fair housing training.

So I'm not trying to get on my high horse on this one and say that we're the end all and be all of fair housing, but the way that housing has been working over the past 50 years can't be the way it works over the next 50. Okay, so you've been in Seattle now for a while, and I know you don't want to badmouth the home team, but you. I love Seattle, but you felt like Vancouver's kind of where the action is, or. Yeah. Well, it's almost a tale of two cities, because, first of all, Vancouver just has different zoning than Seattle. So when you go there, you'll see. It sometimes feels like Hong Kong. It's very dense and bustling, and Seattle has begun to address that. But for a long time, what really shaped the Seattle skyline was this rumor among developers that Jeff Bezos didn't like buildings that were taller than eight or twelve stories. The windows had to open, and it had to be tall enough or low enough that you could take the stairs.

And now, the second part is that Canada is outranking the United States as a destination for international tech talent, that you have all these foreign students who believe in the canadian dream instead. The american dream. And I think it's really becoming this tech center. Some of it's urban planning and some of it's immigration policy. Yeah, but you're gonna. You're hanging tight in Seattle. When you actually think about turning in your american passport. Yeah. And becoming canadian and pumping your gas in liters, it's still a step.

Yeah. No trip to Seattle is really complete without some time on the water. If you're lucky enough, you can find a nerdy sea captain who will take you out with his crew. There you go. Nathan here used to work with the other Nathan at intellectual ventures. Then he started a company called Kimeta that made crazy antennas. And now he started another company named rendered AI. It makes something called synthetic data.

How are you making the first image of this fake x ray? Literally, you can build 3d models in blender or other tools and then simulate them using that. This would just be to make many, many, many x ray images that then get fed into a machine learning system to see what it can spot. You go through the airport, right? Yeah. They, of course, scan your baggage, and that's going through an AI. Of course. You know how they train that AI? They literally have people pack bags and pass them through that so they can be like, that's a knife. That's a banana. That is done by hand today. Okay, so, as you can see, that has some gaps. They just have people spending thousands of hours passing things through. They do.

Okay. Modern machine learning AI needs to go through a process called training, where it's fed tons of data in order to learn how to do a task. Redfin, for example, feeds its algorithms data about real estate. But unlike Redfin, not every company has tons of information at its disposal to train its machine learning algorithms. So what rendered AI does is manufacture data out of thin air. It then feeds AI algorithms with this information to teach them how to perform certain tasks.

These are all synthetically created bags. This is like, add a phone and add a lipstick. You don't just simulate the x ray simulation, right? You can simulate the packing itself. You can simulate the folding of clothes. So we have a lot of different simulations on the platform. So we do satellite imagery, we've done infrared imagery, we've done security cameras. We've actually done microscopy and studied oocytes and human reproduction. So you're making, like, fake cells? Well, yeah, like fake embryos. Sometimes you want to run tests to see if the AI behind an expensive piece of technology will even work.

For example, startups are rushing to send satellites into orbitz that photograph the earth in different ways. They're packed full of expensive, often experimental sensors. Rendered AI can help gauge if these sensors will do the job they're supposed to. Before you pay to build and launch the satellites, you can imagine questions like, hey, I want to detect forest fires, right? Okay. I've got visible cameras in space. How much sooner can I detect a forest fire if I was detecting an AI? So you guys could sort of like, simulate what forest fires would look like from space and do that millions of times. And then maybe we start to see that, like, these patterns kind of emerge before one breaks out.

Exactly. What Nathan is talking about is a weird concept. The AI's need information to become good at their job, and they're now being fed by other AI systems.

Soon, he imagines AI's training the AI's that train the AI's. I mean, what we're doing with synthetic data is about not just training the AI, but training the tools that allow us to do this world building and procedural generation. And it does feed on itself and. Absolutely, like, that's going to lead to all sorts of crazy and unexpected things. Although I will say I'd start by being able to, like, successfully detect cars. Yeah, it'll be really good if we can do that really well. Everybody ready to tack? Yep. All right. Coming about. Perhaps there's a divine entity up there with a giant computer simulating all of this. If so, well done, supreme being. But for the moment, I'm gonna assume this is real and I'm going to revel in it. Next up on hello World, I head to Canada to pound some maple syrup. It's pure. And visit a not so paranoid Android.

Technology, Innovation, Global, Real Estate, Machine Learning, Seattle, Bloomberg Originals